Introduction

People manipulate faces.

Give some examples.

Scope of this type of research.

Common Techniques

Photoshop/Image editors

  • (Gronenschild et al. 2009)

Commerical morphing

  • 831 Google Scholar responses for “fantamorph face”
  • 158 Google Scholar responses for “WinMorph face”
  • Many others: MorphThing (no longer available), xmorph, et. Basically impossible to survey the literature about the methods used because of poor documentaation

Codable Methods

  • imagemagick
  • Matlab
  • Psychomorph
  • WebMorph

Reproducibility!

I gave up on a research project once because I couldn’t figure out how to manipulate spatial scale in MatLab to make my stimuli look like a relevant paper. When I contacted the author, they didn’t know how the stimuli were created because a postdoc just did it in photoshop.

Faces are sampled, so replications should sample new faces as well as new participants.

Difficulty in creating equivalent face stimuli is a barrier to this, resulting in stimulus sets that are used across dozens or hundreds of papers.

  • The Chicago Face Database (“The Chicago Face Database: A Free Stimulus Set of Faces and Norming Data.” 2015) has been cited in almost 800 papers.
  • Ekman POFA selling for $399 for " 110 photographs of facial expressions that have been widely used in cross-cultural studies, and more recently, in neuropsychological research".
  • Image sets are often private and reused without clear attribution (FRL and Perception Lab are particularly bad for this).

Main techniques

Averaging

  • Visualise group differences

Transforming

  • Sexual dimorphism

Methods

Averaging

Transforming

Case Study

London Face Set

Delineation

Automatic versus manual delineation.

Normalisation

Why normalise?

2 point versus Procrustes normalisation (in webmorphR)

lisa <- faces("lisa")

orig <- plot(lisa, pt.plot = TRUE, labels = "", nrow = 1)

twopt <- align(lisa, pt1 = 63, pt2 = 81, patch = TRUE) %>% 
  plot(pt.plot = TRUE, labels = "", nrow = 1)

# any two points that are standard on the image
# should work for procrustes alignment
lisa_proc <- align(lisa, pt1 = 63, pt2 = 81,
                      procrustes = TRUE, patch = TRUE)

procr <- plot(lisa_proc, pt.plot = TRUE, 
              labels = "", nrow = 1)

cowplot::plot_grid(orig, twopt, procr, nrow = 3, 
                   labels = c("original", "two-point", "procrustes"))

Masking

(effect in masc paper)

Averaging

Texture/no

Symmetrising

How this is different from LL/RR mirroring.

Sexual dimorphism transform

Continuum

Self-resemblance transform

Discussion

References

We used R [Version 4.0.2; R Core Team (2020)] and the R-packages papaja [Version 0.1.0.9997; Aust and Barth (2020)], and webmorph [Version 0.0.0.9001; DeBruine (2020)] to produce this manuscript.

Aust, Frederik, and Marius Barth. 2020. papaja: Create APA Manuscripts with R Markdown. https://github.com/crsh/papaja.
DeBruine, Lisa. 2020. Webmorph: Morph Faces. https://github.com/facelab/webmorph.
Gronenschild, Ed H. B. M., Floortje Smeets, Eric F. P. M. Vuurman, Martin P. J. van Boxtel, and Jelle Jolles. 2009. “The Use of Faces as Stimuli in Neuroimaging and Psychological Experiments: A Procedure to Standardize Stimulus Features.” Behavior Research Methods 41: 1053–60. https://doi.org/10.3758/BRM.41.4.1053.
R Core Team. 2020. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/.
“The Chicago Face Database: A Free Stimulus Set of Faces and Norming Data.” 2015. Behavior Research Methods 47: 1122–35. https://doi.org/10.3758/s13428-014-0532-5.